This study provides the most comprehensive Model-Based Total Evidence (MBTE) phylogenetic analyses of the clade Gymnotiformes to date, reappraising relationships using a dataset comprised of six genes (5277bp) and 223 morphological characters, and an ingroup taxon sample including 120 of 212 valid species representing 34 of the 35 extant genera. Our MBTE analyses indicate the two main gymnotiform clades are Gymnotidae and Sternopygoidei, the latter comprised of Rhamphichthyoidea (Rhamphichthyidae+Hypopomidae) and Sinusoidea (Sternopygidae+Apteronotidae). Within Gymnotidae, Electrophorus and Gymnotus are sister taxa, and Gymnotus includes the following six clades: (i) G. pantherinus clade, (ii) G. coatesi clade, (iii) G. anguillaris clade, (iv) G. tigre clade, (v) G. cylindricus clade, and (vi) G. carapo clade. Within Rhamphichthyoidea, Steatogenae (Steatogenys+Hypopygus) is a member of Rhamphichthyidae, and Hypopomidae includes the following clades: (i) Akawaio, (ii) Hypopomus, (iii) Microsternarchini, and (iv) Brachyhypopomus. Within Sternopygidae, Sternopygus and Eigenmanninae are sister groups, Rhabdolichops is the sister to other Eigenmanninae, Archolaemus+Distocyclus is the sister to Eigenmannia, and Japigny is nested within Eigenmannia. Within Apteronotidae, Sternarchorhamphinae (Sternarchorhamphus+Orthosternarchus) is the sister to Apteronotinae, Adontosternarchus is the sister group to other Apteronotinae, Sternarchorhynchini (Sternarchorhynchus+Platyurosternarchus) is the sister to Navajini, and species assigned to Apteronotus are members of two separate clades: (i) A. sensu stricto in the Apteronotini, and (ii) the "A." bonapartii clade in the Navajini.

Neotropical freshwater fishes (NFFs) constitute the most diverse continental vertebrate fauna on Earth, with more than 6,200 named species compressed into an aquatic footprint <0.5% the total regional land-surface area and representing the greatest phenotypic disparity and functional diversity of any continental ichthyofauna. Data from the fossil record and time-calibrated molecular phylogenies indicate that most higher taxa (e.g., genera, families) diversified relatively continuously through the Cenozoic, across broad geographic ranges of the South American platform. Biodiversity data for most NFF clades support a model of continental radiation rather than adaptive radiation, in which speciation occurs mainly in allopatry, and speciation and adaptation are largely decoupled. These radiations occurred under the perennial influence of river capture and sea-level oscillations, which episodically fragmented and merged portions of adjacent river networks. The future of the NFF fauna into the Anthropocene is uncertain, facing numerous threats at local, regional, and continental scales. Expected final online publication date for the Annual Review of Ecology, Evolution, and Systematics, Volume 51 is November 2, 2020. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.

Ultraconserved (UCEs) are popular markers for phylogenomic studies. They are relatively simple to collect from distantly-related organisms, and contain sufficient information to infer relationships at almost all taxonomic levels. Most studies of UCEs use partitioning to account for variation in rates and patterns of molecular evolution among sites, for example by estimating an independent model of molecular evolution for each UCE. However, rates and patterns of molecular evolution vary substantially within as well as between UCEs, suggesting that there may be opportunities to improve how UCEs are partitioned for phylogenetic inference. We propose and evaluate new partitioning methods for phylogenomic studies of UCEs: Sliding-Window Site Characteristics (SWSC), and UCE Site Position (UCESP). The first method uses site characteristics such as entropy, multinomial likelihood, and GC content to generate partitions that account for heterogeneity in rates and patterns of molecular evolution within each UCE. The second method groups together nucleotides that are found in similar physical locations within the UCEs. We examined the new methods with seven published data sets from a variety of taxa. We demonstrate the UCESP method generates partitions that are worse than other strategies used to partition UCE data sets (e.g., one partition per UCE). The SWSC method, particularly when based on site entropies, generates partitions that account for within-UCE heterogeneity and leads to large increases in the model fit. All of the methods, code, and data used in this study, are available from https://github.com/Tagliacollo/PartitionUCE. Simplified code for implementing the best method, the SWSC-EN, is available from https://github.com/Tagliacollo/PFinderUCE-SWSC-EN.